Ultimately marketers are looking to answer one key question with a revenue attribution solution: How should I allocate my marketing budget to get the most growth I can from my scarce marketing dollars?
A hot area addressing a big challenge, vendors have descended on marketers with claims of magic algorithms and 24-hour deployments. Marketers have grown cynical, either from relentless vendors or failed deployments, leading some to dismiss the attribution process entirely. Unfortunately, this is a misguided conclusion. A deep breath and an expectations reset allows the marketer to see that revenue attribution is no different than any other area where they embrace the ability of data-driven decision frameworks to minimize uncertainty and improve outcomes. Successful revenue attribution methodologies typically share the following characteristics:
Broad Organizational Buy-In
All the key stakeholders are at the table providing input: e-commerce, direct mail, retail, and finance. Without buy-in / sign-off from all of the key players, any attribution approach is destined to fail.
All-Inclusive Marketing and Behavioral Data
All the marketing and behavioral data that can be tied to a customer, from all channels is incorporated. If you aren’t capturing it all today or can’t tie it to a customer, don’t wait. Let the initial process reveal opportunities and put in a plan to improve capture, quality, and connectivity over time.
Solid Statistical Foundation
Attribution based purely on rules (first- or last-click, “matchback”, etc.) was historically adequate, but as the number of channels and the velocity of campaigns increased these methods led to inaccurate attribution and budget allocation. Sound statistical approaches algorithmically address this growing complexity, resulting in more accurate attribution and marketing-mix optimization. The best approaches:
- model buying behavior at the order and customer/cookie level,
- account for both campaigns and non-marketing factors (e.g. seasonality, brand loyalty, retail proximity, etc.),
- leverage multiple models to address attribution variances across order channel (phone, web, store) and customer lifecycle (new to file, reactivated, etc.),
- recognize the declining impact of each campaign over time,
- calculate the difference in response rate caused by the number and mix of campaigns and their relative order of occurrence.
Granular Cost Allocation
Many methodologies incorporate costs only at the highest levels (e.g. marketing channel). Allocating marketing costs at the level of orders and their attributed campaigns is what enables the marketer to evaluate the performance of specific emails, keywords, affiliates, etc., providing insights into the types of campaigns that should be emphasized going forward, not just aggregate channel spend.
Appetite for Action
It helps if all order data has been granularly tagged with your attribution learnings. Various aggregations of this low-level data will reveal ROI and testable insights on marketing mix across customer populations, marketing channels, and even campaigns within channels. You may not believe them. They may fly in the face of your intuition. Don’t panic. Perform solid marketing tests to determine if the implications hold water.
Marketers who follow this straightforward approach are realizing significant gains over those who don’t, not because they’ve discovered a secret algorithm, but because they decided to apply the same data driven approach to decision making they have used in other areas their whole careers.